Estimation of soil water storage change from clay shrinkage using satellite radar interferometry

Abstract

Measurements of soil water storage are hard to obtain on scales relevant for water management and policy making. Therefore, this research develops a new measurement methodology for soil water storage estimation in clay containing soils. The proposed methodology relies on the specific property of clay soils to shrink when drying and to swell when (re-)wetted, and the capabilities of a remote sensing technique called satellite based radar interferometry (InSAR) to measure centimetre to millimetre scale displacements of the soil surface. The objective of this thesis was to develop the application of InSAR for soil water storage change estimation on the field scale to regional scale. Two relations are investigated: 1) the relation between water storage change and surface elevation change as a result of swelling and shrinkage of a clayey soil; and 2) the relation between these surface elevation changes and InSAR phase observations.

The shrinkage potential of the soil is very important for successful application of radar interferometry to measure vertical deformation as a result of swelling and shrinkage of clay. Therefore, the shrinkage potential and the water storage change-volume change relation (called the soil shrinkage characteristic, SSC) have been quantified in the laboratory for clay aggregates from the study area in the Purmer, the Netherlands. The clay content of the sampled soil ranged from 3.4 to 23.6%. The aggregates had moderate shrinkage potential over the soil moisture content range from saturation to air-dryness. Shrinkage phases were distinguished based on the portion of water content change that was compensated by volume change. Approximately 40-50% of water was released in the normal shrinkage phase, where loss of water is fully compensated by volume change. However, the residual shrinkage phase, where volume change is smaller than water content change, started at approx. 50% normalized soil moisture content (actual moisture content with respect to the moisture content at saturation).

In case of normal shrinkage, soil water storage change can be directly derived from soil volume change. If additionally, clay shrinkage is isotropic, the soil water storage change can be derived from vertical shrinkage measurements. The range of normal and isotropic shrinkage has been assessed in a drying field soil in the study area. To do so, soil water storage change was derived from soil moisture content sensors and groundwater level, and volume change estimates were obtained from soil layer thickness change measurements by ground anchors. Unlike for the aggregates, normal shrinkage was not observed for the field soil, but rather a large degree of linear (basic) shrinkage was observed. In the upper soil layers in the field, normalized soil moisture content below 50% has been observed when drying out. Based on the aggregate SSC, this indicates the occurrence of residual and zero shrinkage in this situation, resulting in less than normal shrinkage when the total unsaturated zone is considered. The water content change-volume change relation thus depends on the scale considered. It was also found that the relation depends on drying intensity, from comparison between shrinkage in a period with prolonged drying and shrinkage in a period with alternating drying end re-wetting.

For the field soil, volume change larger than soil water storage change was observed when assuming isotropic shrinkage. This unrealistic result made clear that the assumption of isotropic shrinkage is invalid. Therefore a correction of the shrinkage geometry factor rs, including dependence of shrinkage geometry on soil moisture content, has been proposed. This correction yielded rs-values between 1.38 and 3. Dynamics of subsidence porosity (i.e. vertical shrinkage) calculated from the aggregate SSC, and comparison with surface elevation change data from the field study also indicated rs-values smaller than 3. Values of rs below 3, indicate that vertical shrinkage (subsidence) is dominant over horizontal shrinkage (cracking).

Satellite based radar interferometry was applied to measure vertical deformation resulting from clay shrinkage, and evaluate the potential for soil water storage change estimation on the field scale to regional scale. Phase differences between adjacent fields were observed in interferograms over the Purmer area which were hypothesised to be caused by relative motion of the surface level. The combination of a sequence of interferograms covering short time intervals and measurements of soil surface elevation changes in time from ground anchors, indeed revealed similar dynamics in both data. Relative changes between fields in winter were explained by a different effect of frost heave in a bare soil and in a soil permanently covered by grass. Noise in interferograms over agricultural fields was successfully reduced, by multilooking over entire fields. The effect of soil type and land use on phase observation was qualitatively assessed, indicating that agricultural crop fields offer the best phase estimates in winter, while grass fields are more coherent in summer. The results underline the need for careful selection of agricultural fields or areas to base InSAR analysis on.

The differential analysis between fields was extended to time series analysis of phase, to obtain deformation estimates with respect to a stable reference, including correction for unwanted phase contributions and temporal phase unwrapping. The correction of unwanted phase contributions specifically included the soil moisture dielectric effect. This effect was considered by predicting interferometric phase based on in situ measured soil moisture contents. The soil moisture dielectric effect was shown to be much smaller than shrinkage phase in our case study. A simple model was developed to estimate vertical shrinkage, using assumption on shrinkage behaviour (normal and isotropic shrinkage) and an approximation of water storage change from precipitation and evapotranspiration data. Using this model, temporal phase unwrapping results were corrected. The corrections for soil moisture dielectric phase and the correction of phase unwrapping both improved vertical shrinkage measurements from InSAR.

The results in this thesis make clear that vertical clay shrinkage can be estimated from InSAR. At the same time, these results show that clay shrinkage is a considerable phase contribution to interferometric phase and can therefore cause unwrapping and interpretation errors when not accounted for. To estimate vertical clay shrinkage from InSAR, a shrinkage model including assumptions of normal and isotropic shrinkage, proved useful in the phase unwrapping procedure in this case study. However, using the same assumptions to compute water storage change from these InSAR estimates, will in many cases produce inaccurate results. Therefore, in order to use InSAR for estimating soil water storage change in clay soils, the soil shrinkage characteristic, soil moisture dependency of the shrinkage geometry factor, and the effect of variable drying and wetting conditions, need to be considered.